|Publication number||US7796133 B1|
|Application number||US 10/730,965|
|Publication date||Sep 14, 2010|
|Filing date||Dec 8, 2003|
|Priority date||Nov 18, 2002|
|Publication number||10730965, 730965, US 7796133 B1, US 7796133B1, US-B1-7796133, US7796133 B1, US7796133B1|
|Inventors||Mark M. Leather, Eric Demers|
|Original Assignee||Ati Technologies Ulc|
|Export Citation||BiBTeX, EndNote, RefMan|
|Patent Citations (55), Non-Patent Citations (11), Referenced by (3), Classifications (10), Legal Events (3)|
|External Links: USPTO, USPTO Assignment, Espacenet|
This application is a continuation of U.S. application Ser. No. 10/716,946, filed Nov. 18, 2003, now abandoned which claims priority from U.S. provisional application No. 60/427,338, filed Nov. 18, 2002.
1. Field of the Invention
The present invention relates computer graphics. More specifically, one or more embodiments of the present invention relate to a unified shader.
Portions of the disclosure of this patent document contain material that is subject to copyright protection. The copyright owner has no objection to the facsimile reproduction by anyone of the patent document or the patent disclosure as it appears in the Patent and Trademark Office file or records, but otherwise reserves all copyright rights whatsoever.
2. Background Art
Computer systems are often used to generate and display graphics on an output device such as a monitor. Display images are made up of thousands of tiny dots, where each dot is one of thousands or millions of colors. These dots are known as picture elements, or “pixels”. Each pixel has multiple attributes associated with it, including a color and a texture. A shader is used to perform arithmetic and logical operations on all of the attributes of the pixel to achieve one value that is used when the pixel is output. Typically graphics systems use separate shaders for texture and color. Separate shaders make their use difficult and expensive. This problem can be understood by reviewing the way that graphics systems process data.
Each pixel has a color, with the color of each pixel being represented by a number value stored in the computer system. A three dimensional display image, although displayed using a two dimensional array of pixels, may in fact be created by rendering of a plurality of graphical objects. Examples of graphical objects include points, lines, polygons, and three dimensional solid objects. Points, lines, and polygons represent rendering “primitives” which are the basis for most rendering instructions. More complex structures, such as three dimensional objects, are formed from a combination or mesh of such primitives. To display a particular scene, the visible primitives associated with the scene are drawn individually by determining those pixels that fall within the edges of the primitive, and obtaining the attributes of the primitive that correspond to each of those pixels. The obtained attributes are used to determine the displayed color values of applicable pixels.
Sometimes, a three dimensional display image is formed from overlapping primitives or surfaces. A blending function based on an opacity value associated with each pixel of each primitive is used to blend the colors of overlapping surfaces or layers when the top surface is not completely opaque. The final displayed color of an individual pixel may thus be a blend of colors from multiple surfaces or layers.
In some cases, graphical data is rendered by executing instructions from an application that is drawing data to a display. During image rendering, three dimensional data is processed into a two dimensional image suitable for display. The three dimensional image data represents attributes such as color, opacity, texture, depth, and perspective information. The draw commands from a program drawing to the display may include, for example, X and Y coordinates for the vertices of the primitive, as well as some attribute parameters for the primitive (color and depth or “Z” data), and a drawing command. The execution of drawing commands to generate a display image is known as graphics processing.
A shader performs the processing in the graphics pipeline for each pixel. Namely, a shader performs all of the arithmetic and logical operations on all of the attributes of the pixel to combine them and generate a single color value. The single attribute value is placed in a computer's frame buffer and is used when the pixel is output to a display device. Typically graphics systems divide shaders into a texture address unit and a pixel shader. A texture address unit handles texture address calculations. A pixel shader handles color value calculations.
Historically, two types of categories fall into the shader. A first category is the processing of color, which may include processing Z, all of the source attributes having come from interpolation across primitives or a texturing operation. A second category is the operations performed on the texture coordinates themselves. Some prior systems that operate on texture coordinates place all textures into a single pipeline or process multiple texture images in parallel across multiple pipelines. One scheme with a single pipeline uses indirect texturing. Indirect texturing performs texture operations on a pixel (or a number of pixels in parallel) and then uses the texture value to alter the texture coordinates of the pixel if necessary.
Indirect texturing may be used to perform bump mapping, among other things. Bump mapping uses a first texture image with a first set of texture coordinates and a second texture image with a second set of texture coordinates. The first texture image represents a displacement (or bumpiness) that when combined with the texture coordinates of the second texture image displace the texture coordinates which are then sent back through the pipeline. This produces a textured surface that is not totally a smooth light surface, but rather a more textured, bumpy surface, which is more realistic. Another scheme uses a loopback facility to pass the output of the pipeline back to the input of the pipeline to enhance bump mapping operations.
Still, however, conventional shaders are separated into two classes which perform different shading computations. Shading computations performed on colors are typically low precision, 8 bits for instance. Shading computations performed on texture coordinates are typically of higher precision and range because they are addresses into images and must address the image at a sub-texel resolution. So, typically there is a 12 bit number needed for the coordinate of the texture address plus additional bits, three or four for instance, to identify the sub-texel sample location to avoid aliasing effects. Moreover, these bits require an even higher number of bits to achieve accuracy in the range of 15 to 16 bits. Thus, substantially higher precision is needed in a texture coordinate configured shader than in a color shader (twice the precision or more is typical).
For these reasons, color shaders are used for general applications of color while texture coordinate shaders have been limited to only performing high precision bump mapping displacement algorithms. Due to the differing number of bit precisions for the two shaders, each shader uses its own instruction sets, which constrains the generality, flexibility, and overall usefulness of shading algorithms. Moreover, the underlying drivers and computer hardware involved in shading must know whether it is dealing with a color or a texture coordinate, which is disadvantageous.
Another problem associated with multiple, limited shaders is in the structures that must be implemented to perform the shading necessary to render an image. In particular, displacement mapping requires very large structures to hold pixels that are to be operated upon. These structures typically service the pixels in a first-in, first-out manner (FIFO). The FIFOs using bump mapping become huge and inefficient and a large buffer must also be used to store the interpolated results that are to be operated on. The buffer is typically a register and there is a cost associated with synchronizing data between a separate FIFO and register.
The present invention relates to a unified shader. According to one or more embodiments, the functions of a color shader and a texture coordinate shader are combined into a single, unified shader. The unified shader performs both color shading and texture coordinate shading. The unified shader takes a relatively high numerical precision like a texture coordinate shader, but uses the same precision for color shading. The conventional distinction between shading operations (i.e., color texture map and coordinate texture map or color shading operation and texture coordinate operation) is not handled by the use of separate shaders. In the present invention, such distinction is only the intent of the application and the application program interface (API). In this way, any operation, be it for color shading or texture shading, may loop back into the shader and be combined with any other operation.
In one embodiment, color operations are made to a higher precision so that they are the same precision as the texture shading operations. To accomplish this, enhanced scheduling and pipelining of the shading operations are used. In one embodiment, a pipeline in the shader is created for both texture operations and color operations. In the pipelines, a plurality of ALU/SRAM pairs are used to process the shading operations. These ALU/SRAM pairs are synchronized by an internal clock to ensure proper read/write sequences and execution correctness. A scheduled order for texture and color operations is determined and a scheduler is used with both pipelines. The scheduler is used to time the operations correctly, for instance if a texture operation precedes a color operation in a given sequence, then the scheduler will cause the unified shader to wait for input on the color operation pipeline until the texture pipeline completes its operations.
In prior art systems, indirect texture and bump mapping would have required large FIFO memory structures. The unified shader in the present invention handles such mapping without the huge FIFOs and buffer register needed in a conventional texture shader. Instead, the internal clock scheduling mechanism and architecture ALU/SRAM pairs of the unified shader enable it to use only a single memory structure, which may be a FIFO. The FIFO in one embodiment may be placed in SRAM, which is speedier and more efficient than the conventional method of using a FIFO combined with a buffer. Also, because the buffer register is not needed in the FIFO of the unified shader, the cost associated with synchronizing between a buffer register and a FIFO is removed. All these factors result in the reduction in size of the FIFO used in the unified shader. Moreover, in the present invention, the data and operations that are needed to perform shading may be pipelined through the same FIFO. Though SRAM is used in one embodiment, it can be appreciated that other types of equivalent physical memory implementation can be used as well.
In one embodiment, incoming shader codes are partitioned according level of indirection to ensure all ALU and texture instructions pertaining to a given level of indirection are grouped together. The reason for this is that texture operations within the same level are effectively executed in parallel, and so it is important that there be no cross-dependencies. In one embodiment, indirection levels are numbered starting with 0 at the top of the code, and increasing by 1 for each pass through the texture system. An indirection level of 0 means that all ALU operands or texture addresses are derived directly from the input data (rasterized color and texture addresses). Each subsequent level assignment corresponds to the dependency on the number of passes through the texture system (i.e. level N means that all ALU operands or texture addresses are dependent on the result from N pass(es) through the texture system). Control logic within the unified shader uses a plurality of state machines connected serially by a set of small FIFO's to ensure codes separated by levels are executed in the proper sequence.
A register subsystem is used in one embodiment of the present invention to handle the additional complexity created by out-of-order processing of data by the texture unit, ALU, and control logic. In one embodiment, registers 900 within the unified shader are quad or double buffered. Global commands allow the immediate copying of an entire register set. Typically, those registers expected to change often are quad-buffered. Those registers expected to change infrequently are double-buffered.
In another embodiment, a plurality of unified shaders are connected together to process shading operations in parallel. These unified shaders are connected with minimal gluing logics and are synchronized by a universal clock to ensure proper processing.
These and other features, aspects and advantages of the present invention will become better understood with regard to the following description, appended claims and accompanying drawings where:
The invention relates to a unified shader. In the following description, numerous specific details are set forth to provide a more thorough description of embodiments of the invention. It will be apparent, however, to one skilled in the art, that the invention may be practiced without these specific details. In other instances, well known features have not been described in detail so as not to obscure the invention.
One embodiment of a unified shader is shown in the block diagram of
Unified (Pixel) Shader Architecture
Four identical SRAM/ALU pairs 220, 222, 224, and 226 process the data. SRAM space is allocated as each quad is received from the rasterizer. This space is used to store input values and intermediate variables needed by the shader program. This space remains allocated for the time it takes to completely process the quad. Sufficient storage is available such that multiple quads can reside in the shader at the same time at different stages of completion. Though SRAM is used in one embodiment, it can be appreciated that other types of equivalent physical memory implementation can be used as well.
Each SRAM/ALU performs the following operations in one 4-clock cycle:
Due to the time skewing of each SRAM/ALU paid relative to the previous one, it is possible to use a simple mux to multiplex the texture requests from the four SRAM/ALU pairs into a single stream containing one texture request every clock. The resulting texture data is de-multiplexed and written back into the SRAM's. A control logic block 244 generates the SRAM read and write addresses, and issues the ALU instructions for the first SRAM and ALU 220. Each successive SRAM and ALU receives the same addresses and instruction delayed by one clock from the previous one. This enables the addresses and instructions to be lined up with the time-skewed data.
Each SRAM operates within a 4-clock cycle, allowing a single read/write physical SRAM to support four reads and four writes during a given cycle. This design traditionally would have required multi-ported RAM's or register files. The use for each read and write slot within the 4-clock cycle is defined in table 1.
Rasterized Texture Address (tc)
ALU Source Operand A
Rasterized Color (rc)
ALU Source Operand B
Texture Unit Data (td)
ALU Source Operand C
Texture Unit Address (ta)
Shader Code Partitioning
The first step in programming the unified shader of the present invention is partitioning the shader code. This ensures that all ALU and texture instructions pertaining to a given level of indirection are grouped together. The reason for this is that texture operations within the same level are effectively executed in parallel, and so it is important that there be no cross-dependencies.
In one embodiment, indirection levels are numbered starting with 0 at the top of the code, and increasing by 1 for each pass through the texture system. An indirection level of 0 means that all ALU operands or texture addresses are derived directly from the input data (rasterized color and texture addresses). An indirection level of 1 means that all ALU operands or texture addresses are dependent on the result from a single pass through the texture system. An indirection level of 2 means that all ALU operands or texture addresses are dependent on the result from two passes through the texture system.
The control logic for the Unified Pixel Shader of one embodiment of the present invention comprises a number of state machines connected serially by a set of small FIFO's. As data for each block of pixels is received from the rasterizer, a “control token” is generated. The control token contains a small amount of information describing this block of pixels. The control token is passed serially between state machines to indicate who has control over this block of data. Once the control token has reached the last state machine in the chain and the final processed data has been output, the control token is discarded. For a 4-level system, 10 state machines are required. This embodiment's control logic is shown in
Rasterizer 400 generates packets of data containing information for a block of 16 pixels (4 quads). Each pixel contains one or more sets of texture coordinates (texture addresses), and one or more color values. The time needed to transfer this packet is dependent on the number of texture coordinates and colors.
The input state machine 410 receives notification that a packet of data is being received. The input state machine then allocates storage space (for storing input and temporary values) for the block of 16 pixels—each of the four SRAM's will own 4 pixels. The input state machine then instructs the SRAMS's to write the (one or more) sets of texture coordinates and color data to the correct locations within this recently allocated storage space. Upon completion, the input state machine generates a control token for this block of data, and passes it to the level 0 Texture machine 430.
The level 0 texture machine 430 implements the level 0 Texture sequence as shown in
The level 0 ALU machine 450 implements the level 0 ALU sequence as shown in
The level 1 texture machine 455 operates in the same way as the level 0 texture machine 430, passing the control token to the level 1 ALU machine on completion of its section of code. Thus the control token propagates through all 10 machines. Eventually, the command reaches the output state machine. The output state machine instructs the output formatter to pull data from the output FIFO, which is then reformatted and sent out of the Unified Shader to the frame buffer. Following this, the control token is discarded and the storage space for the block of data is deallocated.
Only one ALU state machine may have access to the SRAM address bus or the instruction decode bus at one time. Similarly, only one texture state machine may have access to the SRAM address bus at one time. Arbitration is performed by two arbiter blocks 485 and 486 (one for the ALU state machines and one for the Texture state machines). The arbiters always favor the higher level state machine, preventing half-finished jobs from clogging up the SRAM's.
Different controls token may be resident in the different parts of the control pipelines at the same time, providing an efficient way of hiding latency through the texture unit and maximizing usage of the ALU and the texture units. As an example, the texture unit may be working on a given block while at the same time the ALU may be working on an earlier block.
A comparison of the allocation address pointer for the input state machine, and the deallocation pointer in the output state machine in the chain, gives an indication of how much unallocated SRAM memory is available. When this number falls below a preset mark, the input state machine will stall the rasterizer preventing any new input data.
Where the number of levels of indirection is less than four, certain state machines will go into “pass_through” mode where they immediately copy the control token from the input to the output without performing any work or waiting for arbitration.
Due to the out-of-order processing of data by the texture unit, ALU, and control logic, the normal trickle-down method of loading registers will not work without a complex system of partial flushes that will seriously compromise performance. For this reason, a new register subsystem is used in one embodiment of the present invention and is shown in
On reset, the default quad-buffering tag is set to “A” and the default double-buffered tag is set to “E”. The default tag specifies where mode change registers are written, as well as how subsequent pixel quads are tagged. At the start of the second mode change, a global command is issued to copy register sets “A” to “B” and the default quad-buffered tag is switched to “B”. If the mode change affects double buffered registers, an additional global command is issued to copy register sets “E” to “F” and the default double buffered-tag is switched to F. This continues in rotation with successive mode changes and pixel quads using register sets “C”, “D”, “A”, “B”, etc. for quad buffered registers, and “E”, “F”, “E”, “F”, etc. for double buffered registers.
Prior to issuing a copy command, a check is made of the tag attached to the last pixel quad to leave the shader. If this is equal to the destination register set specified in the copy command, the pipeline is stalled until the situation is rectified. As an example, to copy register set “C” to “D”, one must first look at the data leaving the shader. If this is tagged to use register set “C”, one knows that it cannot be removed from register set “C” without causing an error. Once the tag associated with the data leaving the shader changes to “D”, one can then safely remove the data in “C”, and resume operation.
Multiple unified shaders may be connected (along with a minimum amount of routing and glue logic), to support multiple ALU operations per pixel per clock and multiple independent texture lookups per clock. One embodiment using multiple shaders is shown in
Such a system may be partially disabled (under software control), to allow a fall back to 1 ALU operation per pixel per clock, and 1 texture unit. In the case where the system is partially disabled, the user may choose which of the two unified shaders and which of the two texture units to disable. The system of
The rasterizer 640 outputs two texture addresses and two colors per pixel per clock. Eight quads are rasterized together, with the rasterizer outputting the first set of texture address and color for all eight quads before moving on to the next set. During the “a” cycle, four sets of even texture addresses and colors are passed from the rasterizer to Unified Shader 600 (us_a). During the “b” cycle, the crossbar 650 between the rasterizer and shaders is configured to “swap”, and the next four sets of even texture addresses and colors are passed to unified shader 610 (us_b). Simultaneously, the first four sets of odd texture addresses and colors are passed to unified shader 600 (us_a) after having been delayed by 4 clocks. Thus “us_a” ends up operating on quads 0-3, 8-11, 16-19, etc. and “us_b” ends up operating on quads 4-7, 12-15, 20-23, etc.
When a request for texture owned by the “tx_a” texture unit 620 is called for, this is issued by “us_a” during the “a” cycle for the first set of 4 quads, and by “us_b” during the “b” cycle for the second set of 4 quads. Similarly, when a request for a texture owned by the “tx_b” texture unit 630 is called for, this is issued by “us_a” during the “b” cycle for the first set of quads, and by “us_b” during the “a” cycle for the second set of quads. Since each crossbar is programmed to swap only during the “b” cycle, this ensures that texture units “tx_a” and “tx_b” only process requests for textures which they own. The assignment of textures to the “tx_a” and “tx_b” texture units is done at compile time.
Since shader “us_b” always operates exactly four clocks later than “us_a”, the output data starts coming out of “us_a” for 4 clocks followed by “us_b” for 4 clocks. The outputs are “or'd” together to get a continuous “in order” output stream going to the frame buffer 650. The only modification to the unified shader architecture needed to support an 8-ALU, 2-texture system, is the ability to conditionally issue and receive texture requests during a specific cycle. This requires a small amount of additional complexity to the texture state machine in the control logic. Reverting back to a 4-ALU, 1-texture system involves (a) reconfiguring the crossbars, (b) turning off the conditional texture feature in the unified shaders, and (c) reconfiguring the rasterizer to generate all texture addresses and colors from the tc0 and rc0 ports over a 4-quad cycle.
ALU I/O Description
Each ALU processes one pixel quad (4 pixels) each 4-clock cycle. An example of one embodiment of an ALU is shown in
Even though the result from the ALU operation appears 10 clocks after the instruction is issued, internally the ALU must have a 4-clock latency. The cases where one or more of the source operands are equal to result from the previous of 2nd previous ALU operation, are handled internally by a fast path, and are controlled by the “fastpath” instruction.
Embodiment of Computer Execution Environment (Hardware)
An embodiment of the invention can be implemented as computer software in the form of computer readable program code executed in a general purpose computing environment such as environment 800 illustrated in
Computer 801 may include a communication interface 820 coupled to bus 818. Communication interface 820 provides a two-way data communication coupling via a network link 821 to a local network 822. For example, if communication interface 820 is an integrated services digital network (ISDN) card or a modem, communication interface 820 provides a data communication connection to the corresponding type of telephone line, which comprises part of network link 821. If communication interface 820 is a local area network (LAN) card, communication interface 820 provides a data communication connection via network link 821 to a compatible LAN. Wireless links are also possible. In any such implementation, communication interface 820 sends and receives electrical, electromagnetic or optical signals which carry digital data streams representing various types of information.
Network link 821 typically provides data communication through one or more networks to other data devices. For example, network link 821 may provide a connection through local network 822 to host 823 or to data equipment operated by ISP 824. ISP 824 in turn provides data communication services through the world wide packet data communication network now commonly referred to as the “Internet” 825. Local network 822 and Internet 825 may use electrical, electromagnetic or optical signals which carry digital data streams. The signals through the various networks and the signals on network link 821 and through communication interface 820, which carry the digital data to and from computer 800, are exemplary forms of carrier waves transporting the information.
Processor 813 may reside wholly on client computer 801 or wholly on server 826 or processor 813 may have its computational power distributed between computer 801 and server 826. Server 826 symbolically is represented in
Computer 801 includes a video memory 814, main memory 815 and mass storage 812, all coupled to bi-directional system bus 818 along with keyboard 810, mouse 811 and processor 813. As with processor 813, in various computing environments, main memory 815 and mass storage 812, can reside wholly on server 826 or computer 801, or they may be distributed between the two. Examples of systems where processor 813, main memory 815, and mass storage 812 are distributed between computer 801 and server 826 include the thin-client computing architecture developed by Sun Microsystems, Inc., the palm pilot computing device and other personal digital assistants, Internet ready cellular phones and other Internet computing devices, and in platform independent computing environments, such as those that utilize the Java technologies also developed by Sun Microsystems, Inc.
The mass storage 812 may include both fixed and removable media, such as magnetic, optical or magnetic optical storage systems or any other available mass storage technology. Bus 818 may contain, for example, thirty-two address lines for addressing video memory 814 or main memory 815. The system bus 818 may also include, for example, a 32-bit data bus for transferring data between and among the components, such as processor 813, main memory 815, video memory 814 and mass storage 812. Alternatively, multiplex data/address lines may be used instead of separate data and address lines.
In one embodiment of the invention, the processor 813 is a microprocessor manufactured by Motorola, such as the 680X0 processor or a microprocessor manufactured by Intel, such as the 80X86, or Pentium processor, or a SPARC microprocessor from Sun Microsystems, Inc. However, any other suitable microprocessor or microcomputer may be utilized. Main memory 815 may be comprised of dynamic random access memory (DRAM). Video memory 814 may be a dual-ported video random access memory. One port of the video memory 814 may be coupled to video amplifier 816. The video amplifier 816 may be used to drive a display/output device 817, such as a cathode ray tube (CRT) raster monitor. Video amplifier 816 is well known in the art and may be implemented by any suitable apparatus. This circuitry converts pixel data stored in video memory 814 to a raster signal suitable for use by display/output device 817. Display/output device 817 may be any type of monitor suitable for displaying graphic images.
Computer 801 can send messages and receive data, including program code, through the network(s), network link 821, and communication interface 820. In the Internet example, remote server computer 826 might transmit a requested code for an application program through Internet 825, ISP 824, local network 822 and communication interface 820. The received code may be executed by processor 813 as it is received, and/or stored in mass storage 812, or other non-volatile storage for later execution. In this manner, computer 800 may obtain application code in the form of a carrier wave. Alternatively, remote server computer 826 may execute applications using processor 813, and utilize mass storage 812, and/or video memory 815. The results of the execution at server 826 are then transmitted through Internet 825, ISP 824, local network 822 and communication interface 820. In this example, computer 801 performs only input and output functions.
Application code may be embodied in any form of computer program product. A computer program product comprises a medium configured to store or transport computer readable code, or in which computer readable code may be embedded. Some examples of computer program products are CD-ROM disks, ROM cards, floppy disks, magnetic tapes, computer hard drives, servers on a network, and carrier waves.
The computer systems described above are for example only. An embodiment of the invention may be implemented in any type of computer system or programming or processing environment.
Thus, a unified shader is described in conjunction with one or more specific embodiments. The invention is defined by the claims and their full scope of equivalents.
|Cited Patent||Filing date||Publication date||Applicant||Title|
|US4885703||Nov 4, 1987||Dec 5, 1989||Schlumberger Systems, Inc.||3-D graphics display system using triangle processor pipeline|
|US5179640||Feb 1, 1990||Jan 12, 1993||Oce Graphics Usa Inc.||Generating repeating patterns from a pattern header|
|US5550962||Apr 10, 1995||Aug 27, 1996||Hitachi, Ltd.||System for selectively performing parallel or sequential drawing processing|
|US5745118||Jun 6, 1995||Apr 28, 1998||Hewlett-Packard Company||3D bypass for download of textures|
|US5794016||Dec 11, 1995||Aug 11, 1998||Dynamic Pictures, Inc.||Parallel-processor graphics architecture|
|US5818469||Apr 10, 1997||Oct 6, 1998||International Business Machines Corporation||Graphics interface processing methodology in symmetric multiprocessing or distributed network environments|
|US5905506||Aug 26, 1996||May 18, 1999||Adobe Systems Incorporated||Shared tile image representations|
|US5977997||Mar 6, 1997||Nov 2, 1999||Lsi Logic Corporation||Single chip computer having integrated MPEG and graphical processors|
|US5999196 *||Jul 29, 1998||Dec 7, 1999||Sun Microsystems, Inc.||System and method for data multiplexing within geometry processing units of a three-dimensional graphics accelerator|
|US6118452||Aug 5, 1997||Sep 12, 2000||Hewlett-Packard Company||Fragment visibility pretest system and methodology for improved performance of a graphics system|
|US6184906||Jun 30, 1997||Feb 6, 2001||Ati Technologies, Inc.||Multiple pipeline memory controller for servicing real time data|
|US6219062||Mar 10, 1995||Apr 17, 2001||Hitachi, Ltd.||Three-dimensional graphic display device|
|US6222550||Dec 17, 1998||Apr 24, 2001||Neomagic Corp.||Multiple triangle pixel-pipelines with span-range pixel interlock for processing separate non-overlapping triangles for superscalar 3D graphics engine|
|US6292200||Oct 23, 1998||Sep 18, 2001||Silicon Graphics, Inc.||Apparatus and method for utilizing multiple rendering pipes for a single 3-D display|
|US6323860||Mar 17, 1999||Nov 27, 2001||Nvidia Corporation||Circuit and method for deferring the binding of render states to primitives in a graphics system|
|US6344852||Mar 17, 1999||Feb 5, 2002||Nvidia Corporation||Optimized system and method for binning of graphics data|
|US6353439||Dec 6, 1999||Mar 5, 2002||Nvidia Corporation||System, method and computer program product for a blending operation in a transform module of a computer graphics pipeline|
|US6380935||Mar 17, 1999||Apr 30, 2002||Nvidia Corporation||circuit and method for processing render commands in a tile-based graphics system|
|US6384824 *||Aug 20, 1999||May 7, 2002||Microsoft Corporation||Method, system and computer program product for multi-pass bump-mapping into an environment map|
|US6407736||Jun 18, 1999||Jun 18, 2002||Interval Research Corporation||Deferred scanline conversion architecture|
|US6417858||Dec 23, 1998||Jul 9, 2002||Microsoft Corporation||Processor for geometry transformations and lighting calculations|
|US6424345||Oct 14, 1999||Jul 23, 2002||Ati International Srl||Binsorter triangle insertion optimization|
|US6557083||Jun 30, 2000||Apr 29, 2003||Intel Corporation||Memory system for multiple data types|
|US6570579||Nov 9, 1999||May 27, 2003||Broadcom Corporation||Graphics display system|
|US6573893||Nov 1, 2000||Jun 3, 2003||Hewlett-Packard Development Company, L.P.||Voxel transfer circuit for accelerated volume rendering of a graphics image|
|US6636232||Jan 12, 2001||Oct 21, 2003||Hewlett-Packard Development Company, L.P.||Polygon anti-aliasing with any number of samples on an irregular sample grid using a hierarchical tiler|
|US6650327||Jun 16, 1998||Nov 18, 2003||Silicon Graphics, Inc.||Display system having floating point rasterization and floating point framebuffering|
|US6650330||Jun 21, 2002||Nov 18, 2003||Nvidia Corporation||Graphics system and method for processing multiple independent execution threads|
|US6697063||Nov 25, 1997||Feb 24, 2004||Nvidia U.S. Investment Company||Rendering pipeline|
|US6714203||Mar 19, 2002||Mar 30, 2004||Aechelon Technology, Inc.||Data aware clustered architecture for an image generator|
|US6724394 *||Jun 19, 2001||Apr 20, 2004||Nvidia Corporation||Programmable pixel shading architecture|
|US6731289||May 12, 2000||May 4, 2004||Microsoft Corporation||Extended range pixel display system and method|
|US6750867||Nov 8, 1999||Jun 15, 2004||Imagination Technologies Limited||Image processing apparatus|
|US6753878||Mar 8, 1999||Jun 22, 2004||Hewlett-Packard Development Company, L.P.||Parallel pipelined merge engines|
|US6762763||Jul 1, 1999||Jul 13, 2004||Microsoft Corporation||Computer system having a distributed texture memory architecture|
|US6778177||Apr 10, 2000||Aug 17, 2004||Sp3D Chip Design Gmbh||Method for rasterizing a graphics basic component|
|US6791559||Feb 28, 2002||Sep 14, 2004||3Dlabs Inc., Ltd||Parameter circular buffers|
|US6801203||Dec 22, 1999||Oct 5, 2004||Microsoft Corporation||Efficient graphics pipeline with a pixel cache and data pre-fetching|
|US6809732||Dec 13, 2002||Oct 26, 2004||Nvidia Corporation||Method and apparatus for generation of programmable shader configuration information from state-based control information and program instructions|
|US6864893 *||Nov 22, 2002||Mar 8, 2005||Nvidia Corporation||Method and apparatus for modifying depth values using pixel programs|
|US6864896||May 15, 2001||Mar 8, 2005||Rambus Inc.||Scalable unified memory architecture|
|US6897871||Nov 20, 2003||May 24, 2005||Ati Technologies Inc.||Graphics processing architecture employing a unified shader|
|US6980209 *||Jun 14, 2002||Dec 27, 2005||Nvidia Corporation||Method and system for scalable, dataflow-based, programmable processing of graphics data|
|US7015913||Jun 27, 2003||Mar 21, 2006||Nvidia Corporation||Method and apparatus for multithreaded processing of data in a programmable graphics processor|
|US7061495||Dec 8, 2003||Jun 13, 2006||Ati Technologies, Inc.||Method and apparatus for rasterizer interpolation|
|US7170515||Dec 5, 2003||Jan 30, 2007||Nvidia Corporation||Rendering pipeline|
|US20020145612||Jan 29, 2002||Oct 10, 2002||Blythe David R.||Method and system for minimizing an amount of data needed to test data against subarea boundaries in spatially composited digital video|
|US20030076320 *||Oct 18, 2001||Apr 24, 2003||David Collodi||Programmable per-pixel shader with lighting support|
|US20030164830||Mar 1, 2002||Sep 4, 2003||3Dlabs Inc., Ltd.||Yield enhancement of complex chips|
|US20040041814||Aug 30, 2002||Mar 4, 2004||Wyatt David A.||Method and apparatus for synchronizing processing of multiple asynchronous client queues on a graphics controller device|
|US20040100471||Jun 12, 2003||May 27, 2004||Leather Mark M.||Dividing work among multiple graphics pipelines using a super-tiling technique|
|US20040164987||Feb 24, 2003||Aug 26, 2004||Microsoft Corporation||Usage semantics|
|US20050068325||Sep 29, 2003||Mar 31, 2005||Ati Technologies, Inc.||Multi-thread graphic processing system|
|US20050200629||Apr 29, 2005||Sep 15, 2005||Steven Morein||Graphics processing architecture employing a unified shader|
|US20060170690||Feb 28, 2006||Aug 3, 2006||Ati Technologies, Inc.||Method and apparatus for rasterizer interpolation|
|1||Akeley K. et al., "High-Performance Polygon Rendering", ACM Computer Graphics vol. 22 No. 4; Aug. 1988, pp. 239-246.|
|2||Breternitz, Jr., Mauricio et al.; Compilation, Architectural Support, and Evaluation of SIMD Graphics Pipeline Programs on a General-Purpose CPU; IEEE; 2003; pp. 1-11.|
|3||Crockett, Thomas W.; An introduction to parallel rendering; Elsevier Science B.V.; 1997; pp. 819-843.|
|4||Elias, Hugo; "Polygon Scan Converting"; from http://freespace/virgin.net/hugo.elias/graphics/x-polysc.htm; dated Jul. 26, 2005; pp. 1-7.|
|5||Elias, Hugo; "Polygon Scan Converting"; from http://freespace/virgin.net/hugo.elias/graphics/x—polysc.htm; dated Jul. 26, 2005; pp. 1-7.|
|6||European Search Report from European Patent Office; European Application No. 03257464.2; dated Jan. 4, 2006.|
|7||Foley, James et al.; Computer Graphics, Principles and Practice; Addison-Wesley Publishing Company; 1990; pp. 873-899.|
|8||Fuchs, Henry et al., "Pixel-Planes 5: A Heterogeneous Multiprocessor Graphics System Using Processor-Enhanced Memories"; Computer Graphics, vol. 23, No. 3, Jul. 1989; pp. 79-88.|
|9||Humphreys, Greg et al.; WireGL: A Scalable Grpahics System for Clusters; ACM Siggraph; Aug. 2001; pp. 129-140.|
|10||International Search Report for PCT Patent Application PCT/IB2004/003821 dated Mar. 22, 2005.|
|11||Montrym, John S. et al.; InfiniteReality: A Real-Time Graphics System; Silicon Graphics Computer Systems; 1997; pp. 293-302.|
|Citing Patent||Filing date||Publication date||Applicant||Title|
|US9135017 *||Jan 16, 2008||Sep 15, 2015||Ati Technologies Ulc||Configurable shader ALU units|
|US20080189524 *||Jan 16, 2008||Aug 7, 2008||Elaine Poon||Configurable shader ALU units|
|US20090051687 *||Oct 24, 2006||Feb 26, 2009||Mitsubishi Electric Corporation||Image processing device|
|U.S. Classification||345/426, 345/589, 345/552, 345/582|
|International Classification||G06T11/40, G09G5/02, G06T15/50, G09G5/00|
|Mar 2, 2004||AS||Assignment|
Owner name: ATI TECHNOLOGIES INC., CANADA
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNORS:LEATHER, MARK M.;DEMERS, ERIC;REEL/FRAME:014391/0276
Effective date: 20040122
|Feb 12, 2014||FPAY||Fee payment|
Year of fee payment: 4
|Jan 18, 2017||AS||Assignment|
Owner name: ATI TECHNOLOGIES ULC, CANADA
Free format text: CHANGE OF NAME;ASSIGNOR:ATI TECHNOLOGIES INC.;REEL/FRAME:041343/0208
Effective date: 20061025